pub / newsplanet

commit dde744b5002b48c248ccfcb7f208616ed94ef1e5
parent 47766b32d7c6edb18c90c0fcef6c3aa54e1dc39f
Author: JayVii <jayvii[AT]posteo[DOT]de>
Date:   Sat, 15 Mar 2025 12:08:13 +0100

clean up HTML more rigorously

Diffstat:
M
run.sh
 | 
18
+++++++++++-------
M
template.tt
 | 
6
+++---

2 files changed, 14 insertions(+), 10 deletions(-)
diff --git a/run.sh b/run.sh
@@ -2,7 +2,7 @@
 
 # fetch input config file, fall back to "centre"
 if [ -z $1 ]; then
-  CONFIG="01_pulse.yaml"
+  CONFIG="pulse.yaml"
 else
   CONFIG="$1"
 fi
@@ -25,9 +25,13 @@ printf "Removing images and iframes...\n"
 sed -E -e 's/<img[^>]*>//g' -e 's/<iframe[^>]*>//g' \
   -e 's/<figure[^>]*>//g' -e 's/<\/figure>//g' -i "$HTML"
 
-# remove linebreaks and empty paragraphs
+# remove linebreaks and empty paragraphs as well as inline-links ("more...")
 printf "Clean up HTML...\n"
-sed -E -e 's/<br[^a-z]*[^>]*>//g' -e 's/<p[^a-z]*[^>]*><\/p>//g' -i "$HTML"
+sed -E \
+  -e 's/<br[^a-z]*[^>]*>//g' \
+  -e 's/<p[^a-z]*[^>]*><\/p>//g' \
+  -e 's/<a\ [^<]+<\/a><\!--end-->/<\!--end-->/g' \
+  -i "$HTML"
 
 # insert link to rss/xml file
 printf "Inserting RSS feed file...\n"
@@ -49,19 +53,19 @@ sed -E \
 
 # Insert feeds list
 printf "Inserting feeds list...\n"
-FEEDS=""
+FEEDS="<ul>"
 for i in `seq 1 1 $NFEEDS`; do
   TITLE=`echo "$TITLES" | sed -n "${i},${i}p"`
   WEB=`echo "$WEBS" | sed -n "${i},${i}p"`
-  FEEDS="${FEEDS}<a href=\"${WEB}\">${TITLE}</a><br>"
+  FEEDS="${FEEDS}<li><a href=\"${WEB}\">${TITLE}</a></li>"
 done
-FEEDS=`echo "$FEEDS" | sed -e 's/\//\\\\\//g' -ze 's/\n//g'`
+FEEDS=`echo "${FEEDS}</ul>" | sed -e 's/\//\\\\\//g' -ze 's/\n//g'`
 sed -E -e "s/<\!--FEEDS-->/$FEEDS/" -i "$HTML"
 
 # Insert Update time
 printf "Inserting update time...\n"
 DATE=`date +%c`
-sed -E -e "s/<\!--UPDATED-->/$DATE/" -i "$HTML"
+sed -E -e "s/<\!--UPDATED-->/&#128472; ${DATE}/" -i "$HTML"
 
 # Re-download favicons
 # ./fetch_favicon.sh "$CONFIG"
diff --git a/template.tt b/template.tt
@@ -50,7 +50,7 @@
           This is updated by find-and-replace
           WORKAROUND: [% feed.modified %] is hardcoded to UTC
         -->
-        <p>Last Update:
+        <p>
           <!--UPDATED-->
         </p>
 
@@ -72,9 +72,9 @@
           </h3>
           <div class="article_content">
             [% IF entry.summary.body %]
-              <p>[% entry.summary.body  %]</p>
+              <p><!--start-->[% entry.summary.body  %]<!--end--></p>
             [% ELSE %]
-                [% entry.content.body %]
+                <!--start-->[% entry.content.body %]<!--end-->
             [% END %]
           </div>